×

Knowledge representation and relation nets. (English) Zbl 0928.68105

The Kluwer International Series in Engineering and Computer Science. 506. Boston: Kluwer Academic Publishers. x, 279 p. (1999).
The book introduces a new approach to knowledge representation based on the notion of a relation net. A relation net is defined as a labeled directed graph that represents a finite collection of finite relations of various arities greater or equal 2 over a finite set. The authors claim that this notion overcomes some of the constraints inherent in structural modeling based on ordinary directed graphs and thus broadens the scope of structural modeling. The book consists of three parts. After a brief overview of some of the existing approaches to knowledge representation, the authors introduce in Part 1 the notion of a relation net. A special kind of relation nets, called Concept-Name-Relationship Nets (CNR-nets), and a new knowledge representation schema, called Concept-Relationship Knowledge Structure (CRKS) are defined. An example of a CNR-net is given to show the context-sensitivity of this representation. An extended example of a CRKS is presented in Part 2. It is shown that every path in the CRKS is a derivation path. Part 3 introduces the theory of relation nets. It is shown that relation nets display a new aspect of vertex vulnerability, which does not occur for directed graphs. This leads to a number of new theorems about the properties of relation nets. Directed graphs and binary networks are shown to be subclasses of the classes of relation nets.

MSC:

68T30 Knowledge representation
68-01 Introductory exposition (textbooks, tutorial papers, etc.) pertaining to computer science
PDFBibTeX XMLCite